-
Notifications
You must be signed in to change notification settings - Fork 2
10. fMRI data preprocessing
These are procedures to 'unpack' DICOM images into NIFTI (nii) format. The scripts were tested at /space/maki7/users/fhlin/smsini_array_nccu/063019/unpack on July 7 2019.
Unpacked DICOM images are at three folders:
/space/maki7/users/fhlin/smsini_array_nccu/063019/unpack/bold/005
/space/maki7/users/fhlin/smsini_array_nccu/063019/unpack/bold/007
/space/maki7/users/fhlin/smsini_array_nccu/063019/unpack/bold/009
Two files are needed for FsFast pre-processing. They are session directory (sessdir) and session folder (sessid) files. The content for session directory (sessdir) is
/space/maki7/users/fhlin/smsini_array_nccu/063019
THe content for session folder (sessid) is
unpack
rolle10:/space/maki7/users/fhlin/smsini_array_nccu/063019/unpack> mktemplate-sess -sf sessid -df sessdir
...
...
writing to /autofs/space/maki7/fhlin/smsini_array_nccu/063019/unpack/bold/009/template.nii.gz...
Sun Jul 7 23:05:55 EDT 2019
mktemplate-sess completed
rolle10:/space/maki7/users/fhlin/smsini_array_nccu/063019/unpack> mc-sess -sf sessid -df sessdir -per-run
...
...
Writing to /autofs/space/maki7/fhlin/smsini_array_nccu/063019/unpack/bold/009/tmpdir.mcdat2extreg.141447/pca.nii
mcdat2mcextreg done
Sun Jul 7 23:10:40 EDT 2019
mc-sess completed SUCCESSFULLY
New zipped fMRI images named "fmcpr.nii.gz" will be created in each folder.
For multi-slice sequences, the slice timing needs to be corrected for different slices such that they are synchronized. Here we specified that slice timing correction is applied to motion-corrected images (fmcpr.nii). The slice order is specified to be "Siemens" (-so option).
rolle10:/space/maki7/users/fhlin/smsini_array_nccu/063019/unpack> stc-sess -sf sessid -df sessdir -i fmcpr -o fmcprstc -so siemens
...
...
writing to /autofs/space/maki7/fhlin/smsini_array_nccu/063019/unpack/bold/009/fmcprstc.nii.gz...
stc.fsl Done
Started at Sun Jul 7 23:12:49 EDT 2019
Ended at Sun Jul 7 23:16:02 EDT 2019
stc-sess Done
New zipped fMRI images named "fmcprstc.nii.gz" will be created in each folder.
Smoothing fMRI images can improve the sensitivity if the smoothing kernel matches the fMRI activation size. Here is the script to do so by applying a Gaussian smoothing kernel with full-width-half-maximum (FWHM) of 8 mm.
rolle10:/space/maki7/users/fhlin/smsini_array_nccu/063019/unpack> spatialsmooth-sess -sf sessid -df sessdir -i fmcprstc -o sfmcprstc -outfmt nii -fwhm 8 -no-mask
...
...
Saving to 009/sfmcprstc.nii
SmoothOnly requested, so exiting now
Started at Sun Jul 7 23:21:40 EDT 2019
Ended at Sun Jul 7 23:23:44 EDT 2019
spatiallysmooth-sess Done
New fMRI images named "sfmcprstc.nii" will be created in each folder.
This is the optional step for surface-based fMRI analysis. The EPI files after pre-processing can be "painted" over a cortical surface such that subsequent analysis will be performed in a 2D space.
[This script] calls the fslregister
in FSL to perform registration. The output is a registration file named `register.dat', including parameters for the linear affine transformation.
[This script] calls mri_vol2surf
to convert the values in each EPI volumes to values in a cortical surface. The output file format is .stc
format, which is essentially a 2D matrix of [# of cortical surface] x [# of time points].
[This script] calls mri_vol2vol
to morph EPI volumes in the native space of an individual to EPI volumes in the template, MNI305, space. The output will be MGH files with 2 mm isotropic resolution (76 x 76 x 93 voxels in coronal slices).
The underlay image for morphed EPI data can be accesssed at the mni305.cor.subfov2.mgz
file in FreeSurfer. At MAC, it is /Applications/freesurfer/average/mni305.cor.subfov2.mgz
.